Mortality behavior at advanced ages is a key for understanding the future trends in human longevity. In spite of the extensive literature on the topic, the pattern of mortality at later ages is still a matter for debate, in part because of age misreporting and scarce data at extreme old ages. The problem is exacerbated by the lack of new systematic studies of U.S. old-age mortality, which results in significant disagreement between U.S. governmental agencies and actuarial organizations on the U.S. mortality patterns at advanced ages. This project intends to fill an existing knowledge gap by obtaining improved estimates of old-age mortality trajectories for selected U.S. birth cohorts, by analyzing historical evolution of U.S. cohort mortality after age 80 years, and by testing hypotheses on mortality compression and mortality convergence at advanced ages. To this aim, we will reconstruct old-age mortality for two single-year U.S. birth cohorts (1900 and 1901) by assessing quality of age reporting and by conducting age validation procedures through the linkage of the Social Security Administration Death Master File records to historical resources. The 1900 and 1901 birth cohorts for the U.S. present a unique opportunity to analyze the mortality at advanced ages with a high level of confidence in the results. Advanced ages can be easily verified as these cohorts were too young to qualify for Medicare when it was first introduced, so proof of age was required when they qualified in the following years. The project will also study mortality trajectories and mortality convergence at advanced ages by gender, race, education, and by major causes of death using mortality records from the National Vital Statistics System. These studies will provide us with better outlook on the past and future trends of longevity in the United States.